Skip to content
Data, stimuli, materials, and code for Tarantola et al, Prior preferences beneficially influence social and non-social learning. Nature Communications (2017)
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
analysis_code Delete RT model prediction pickles for size Jul 12, 2017
data
ethics
figures
materials
task_code
.gitignore
README.md

README.md

Prior preferences beneficially influence social and non-social learning

Data, code, and materials for Tarantola, Kumaran, Dayan, & De Martino. Prior preferences beneficially influence social and non-social learning. Nature Communications (2017). Stan fit objects and response time model predictions are too big for GitHub, but are available on request (contact tor.tarantola@gmail.com). They can also be run locally using the Stan scripts (Stan models; see Analysis below) and the script in the "analyses" Jupyter notebook in the analysis_code folder (response time model predictions). Running models locally will result in slightly different outputs due to the stochastic nature of the sampling algorithm.

Organization

  • Data Raw and processed data files are contained in the 'data/' folder. Data from the social experiment are in the 'social' subfolder, and data for the non-social experiment are in the 'non_social' folder. Pilot data are in the 'social_pilot' folder. The 'combined' folder contains the processed data from both the social and nonsocial experiments, combined into a single data file. The participant log includes details on each participant, along with contemporaneous notes on the testing session.
  • Analysis 'analysis_code/' contains IPython notebooks showing the data analysis and simulations included in the manuscript, as well as the Stan models and related scripts ('analysis_code/stan'). Each model is sampled by its matching 'sample_*.py' script.
  • Task Code The PsychoPy task scripts and stimuli are in the 'task_code/' folder. Participants in the social experiment completed 'infer_design1.py' and participants in the nonsocial experiment completed 'infer_design1_no_eyetracking.py.' (Participants in the social group underwent eye tracking to pilot a different study.) Item pairs were generated randomly by the 'pair_generator.py' script, and underlying choices for the practice block were generated by the 'inference_practice_choice_generator.py' script. These pairs and choices were kept the same for all non-pilot participants.

Authors

Tor Tarantola (Department of Psychology, University of Cambridge)
Dharshan Kumaran (Institute of Cognitive Neuroscience, University College London)
Peter Dayan (Gatsby Computational Neuroscience Unit, University College London)
Benedetto De Martino (Institute of Cognitive Neuroscience, University College London)

Contact

tor.tarantola@gmail.com

Figshare

A copy of this repository is also on Figshare (DOI: 10.6084/m9.figshare.5198572), versioned to the git release tags.

You can’t perform that action at this time.